Midjourney Prompt Generator

The main subject is the most important part. Clearly describe what you want to see. Everything else - style, lighting, colors - builds on that. The more specific your description, the better the result.

Parameters control various aspects: --ar for aspect ratio, --chaos for randomness (0-100), --stylize for artistic interpretation (0-1000), --q for quality, and --v for version. These are added at the end of the prompt.

--chaos (0-100) controls the variation between the 4 generated images. --stylize (0-1000) determines how artistically Midjourney interprets the prompt. Low values = more literal, high values = more artistic.

--no is perfect for excluding unwanted elements. For example: '--no people' removes people, '--no text' prevents text in the image. This is often more effective than describing what you don't want.

Use Character References with --cref [image URL]. Describe the character in detail in the first prompt, save the best result and use it as a reference for more images with the same character.

V6 is the latest version with improved prompt accuracy and details. V5.2 was the long-standing standard. Niji is specifically optimized for anime/manga. Raw mode (--style raw) gives you more control with less artistic interpretation.